The Biosynthetic Basis of Cell Size Control

Abstract

Cell size is an important physiological trait that sets the scale of all biosynthetic processes. Although physiological studies have revealed that cells actively regulate their size, the molecular mechanisms underlying this regulation have remained unclear. Here we review recent progress in identifying the molecular mechanisms of cell size control. We focus on budding yeast, where cell growth dilutes a cell cycle inhibitor to couple growth and division. We discuss a new model for size control based on the titration of activator and inhibitor molecules whose synthesis rates are differentially dependent on cell size.

Introduction

Cell size determines the geometry of all intracellular compartments and sets the scale of biosynthetic processes [1,2] (Figure 1). Biosynthesis increases in larger cells, which have proportionally more protein [3–6], total RNA [3,7], total mRNA [8], and mRNA for specific investigated genes [7,9,10]. The proportionate increase in building blocks in larger cells is likely to govern the size of organelles such as the nucleus [11,12], mitochondria [13], centrosome [14], vacuole [15], nucleolus [16], and mitotic spindle [17–21]. Due to its important role in cellular processes, many organisms actively control cell size by coupling growth and division [22–24] and target cell size is often modulated by environmental conditions [23]. Yet, despite the fundamental importance of cell size control, the molecular mechanisms underpinning this regulation have remained elusive.

Figure 1. The Size or Number of Many Cellular Components Increases with Cell Size.

Most proteins and mRNAs increase in direct proportion to cell size so that molecular concentrations are constant in growing cells. Moreover, organelle size often scales with cell size. By contrast, cells of different size often have the same amount of genomic DNA. Recently, it was shown that some proteins, including the cell cycle inhibitor Whi5, are not synthesized in proportion to cell size, to generate size-dependent concentrations and couple growth to division.

Actively controlling cell size requires the generation of a size-dependent biochemical signal. However, biochemical reaction rates are typically determined by the concentrations of the reacting molecules and most protein amounts are proportional to cell size, such that their concentrations are unaffected by cell growth. This raises the question of how a cell can generate size-dependent biochemical signals using constant-concentration proteins. One proposed general mechanism to generate a size-dependent signal from proteins whose concentration is size independent is to localize those proteins in gradients with characteristic length scales [25–27]. The size of the cell can then be measured using the gradient length as a ruler. Specifically, in fission yeast cell size control was proposed to rely on a spatial gradient of Pom1, an inhibitor of mitotic entry localized in a gradient from the cell poles. As cells grow, the concentration of Pom1 at the middle of the cell decreases so that division is triggered [25,26]. However, recent work has cast doubt on this model because Pom1 is not essential for size control [28] and a new model based on increasing local concentration of Cdr2, a kinase activating mitotic entry, has been proposed [29]. Importantly, geometry-based size control mechanisms require a consistent geometry and therefore are unlikely to be employed by animal cells, which are more irregularly shaped. Moreover, geometric mechanisms are unlikely to work well in near-spherical budding yeast where characteristic lengths scale as the volume to the one-third power. A second general mechanism to create a size-dependent signal relies on the localization of proteins, whose amount is proportional to cell size, into compartments whose volume is not proportional to cell size. This gives rise to a size-dependent local concentration that can be used to trigger a cellular transition.

Here we discuss a new general alternative. While most macromolecules accumulate in proportion to cell size, some do not. This differential size dependency of synthesis can be used to generate size-dependent changes in the relative concentrations of activator and inhibitor molecules to couple cell size with specific cellular transitions.

Differential Cell Size Dependence of Protein Synthesis Controls Budding Yeast Size

While geometric gradient-based mechanisms are unlikely to apply to budding yeast, compartmentalization-based mechanisms are a viable alternative. In general, such mechanisms rely on generating a local size-dependent protein concentration by concentrating a protein whose amount is proportional to cell size in a compartment whose volume increases more slowly than the cell volume. We note here that we generally use the words cell size, volume, and total protein content interchangeably because of the small density differences associated with cell cycle progression in budding yeast [30].

In budding yeast, the duration of the G1 phase, between cell division and DNA replication, is strongly affected by cell size, while the duration of S/G2/M depends only weakly on cell size [23,31–35]. An analysis of cell-to-cell variability in G1 duration showed that differences in cell size at birth account for approximately half of the G1 variability, while molecular noise accounts for the other half [34]. The G1/S transition is promoted by the G1 cyclin Cln3 in complex with the cyclin-dependent kinase Cdk1 and by Bck2 through a currently unknown mechanism (cln3Δbck2Δ cells are unviable) [36–38]. Here we focus on the mechanism through which the Cln3 cell cycle pathway promotes division.

Although average cell size in a population is sensitive to CLN3 dose, the concentration of Cln3 protein in G1 does not change appreciably as cells grow [35,39]. This has raised the question of how this putative trigger protein could initiate the cell cycle in a size-dependent manner. It was first proposed that Cln3’s nuclear localization could generate a size-dependent signal. If the ratio of nuclear to cytoplasmic volume decreased as cells grew, this would concentrate Cln3 protein in the nucleus [40]. However, the size of the budding yeast nucleus, as for many other organelles, increases in direct proportion to cell size, so the nuclear concentration of Cln3 would also remain constant [11]. Next, it was suggested that the genome itself, which is a fixed length through G1, might serve as a compartment to concentrate Cln3. In this model, tight binding of Cln3 would result in a size-dependent increase in its local concentration at target sites on promoter regions of genes expressed at the G1/S transition. Thus, the genome would serve as a ruler to count the number of Cln3 molecules, which is directly proportional to cell size [41]. However, cellular DNA concentration per se was recently shown not to affect G1 progression independent of the concentration of Cln3 and its primary target Whi5 [35]. Another model proposes that Cln3 is retained at the endoplasmic reticulum and then released after the growth-dependent accumulation of its chaperone Ydj1 [42,43]. However, we currently lack convincing evidence supporting rapid mid-G1 translocation of wild-type (WT) Cln3 and mutant alleles visible with fluorescence imaging do not show this behavior [35]. Thus, while attractive, compartmentalization-based mechanisms had so far failed to explain size-dependent progression through G1 in budding yeast.

While much effort has focused on how the constant Cln3 concentration could be converted to a size-dependent signal, this is not the only way growth could drive division. A biochemical mechanism for measuring amount rather than concentration is necessary only if all G1/S regulatory proteins exist at constant concentration through G1, an assumption that was never systematically tested. When this assumption was tested [35], it was found that most regulators, including Cln3, are maintained at a constant concentration during G1. However, growth diluted Whi5, which is the downstream target of Cln3 and an inhibitor of cell cycle progression [44–46]. This finding immediately led to a new size control model. Since very little Whi5 is synthesized during G1 [35,47], cell growth could be linked to division through the dilution of this cell cycle inhibitor [35].

Inhibitor-dilution mechanisms will provide growth requirements for division but do not necessarily result in the differential growth requirements for larger-born and smaller-born cells that effect cell size control. An additional mechanism must ensure that smaller-born daughter cells grow more in G1 than larger-born daughter cells. This additional mechanism also works through Whi5, as it was found that larger cells are born with lower Whi5 concentrations [35]. Since cell cycle entry depends on Whi5 concentration, large cells born with a lower concentration require less growth during G1. This results in cell size control by ensuring that larger-born cells grow less in G1 than smaller-born cells. We note that the more detailed version of the inhibitor-dilution model is a stochastic model, which accounts for signicant cell-to-cell variability in G1 duration. The stochastic version of the Whi5-dilution model predicts the rate at which cells enter the division cycle, which is a decreasing function of their Whi5 concentration [35].

That differential growth requirements for larger- and smaller-born cells arise from different Whi5 concentrations at birth raises yet another question. Why are larger and smaller cells born with approximately the same number of Whi5 molecules? Whi5 is a stable protein expressed almost exclusively during the S/G2/M phases of the cell cycle. In sharp contrast to other proteins like Cln3, whose synthesis is proportional to cell size, the rate of Whi5 synthesis is largely independent of cell size [35]. Since S/G2/M duration depends only weakly on cell size [31,33,35] this results in large and small mother cells producing roughly the same amount of Whi5, which is then inherited by the daughter cells. Thus, at its most fundamental level, budding yeast size control results from the differential size dependence of the synthesis of the cell cycle activator Cln3, which scales with cell size, and the inhibitor Whi5, which does not (Figure 2A).

Figure 2. To Link Cell Size with Division, Cells Employ Regulatory Molecules with Cell Size-Dependent Concentrations.

(A) In budding yeast, cell size control is based on the differential synthesis of the cell cycle activator Cln3 and the inhibitor Whi5. While Cln3 is produced in proportion to cell size so that its concentration is constant, Whi5 is produced at a rate independent of cell size so that its concentration is smaller in larger cells. This promotes cell cycle entry in larger cells. (B) In early frog embryos, a similar mechanism senses cell size to control the timing of the mid-blastula transition (MBT) at the 12th division cycle. Histones inhibit the MBT and are at constant concentration while genomic DNA promotes the MBT. DNA concentration doubles at each cell division because in the early frog embryo cells divide without growing. The decreasing histone-to-DNA ratio can then measure cell size to control the timing of the MBT.

Differential Cell Size Dependence of Histone and DNA Concentrations Controls the Frog Mid-Blastula Transition (MBT)

While molecule amounts are typically proportional to cell size, the ability of cells to break this proportionality for specific signaling molecules allows them to create a cell size-dependent signal. Importantly, this approach offers a generally applicable mechanism for cells to sense their size that does not rely on cell geometry, which is highly complex for many eukaryotic cells. A similar mechanism controls the MBT of early frog development (Figure 2B).

During early development frog embryos undergo a specific number of rapid, synchronous, cell divisions before the onset of a coordinated set of events including zygotic transcription, cell motility, and cell cycle lengthening that are collectively known as the MBT [48,49]. During this period, cells do not grow, so that each cell division cycle decreases cell size by half. A longstanding model proposed that, to correctly time the MBT, cells sense the decreasing ratio of cytoplasmic volume to DNA content [50]. In this model, a transcriptional inhibitor at constant concentration would be titrated out by the exponentially increasing DNA concentration.

Recently, histones H3 and H4 were identified as the inhibitory factor titrated against the increasing amount of DNA until a certain threshold is reached to trigger the MBT [51]. In essence, cells sense the histone-to-DNA ratio and generate a size-dependent signal through the fact that DNA concentration is inversely proportional to cell size during the rapid pre-MBT divisions without cell growth. Thus, much like in yeast, differential size dependence of the synthesis of an activator and an inhibitor links cell size to a specific cellular transition.

Differential Size Dependence of Biosynthesis

As highlighted by our two examples above, cells are able to couple their size with a transition by creating distinct cell size dependencies for the synthesis of regulatory molecules. While it is well known that most cellular components, including both organelles and molecules, are produced in proportion to cell size, we do not know why this is the general case. Perhaps unsurprisingly, we know even less about the more exceptional cases where the synthesis of cellular components is not proportional. Although much has been learned about the checkpoint mechanisms ensuring that the genome is replicated precisely once per round of cell division independently of cell size, it seems unlikely that such checkpoints, or feedback mechanisms, are responsible for size-independent protein synthesis. In the case of Whi5, adding an extra copy of the WHI5 gene doubles its rate of synthesis, which again remains independent of cell size [35]. That size-dependent protein concentrations do not rely on feedback mechanisms is consistent with the more general observation that there is little gene dosage compensation in hemizygous diploid[52] and aneuploid [53] yeast cells.

Size-Dependent Biosynthesis

Proceeding from the principle that the more general phenomenon should be understood first, we begin our discussion with what is known about how cells are able to maintain most proteins at constant concentration. We note that we are here concerned only with understanding how cell size affects biosynthesis under a constant environmental condition. The more complex question of how nutrient availability and metabolism regulate biosynthesis, cell growth, and size control are beyond the scope of this review but have been considered elsewhere [23,54–58].

The first step in gene expression is transcription. As for protein amounts, mRNA amounts increase with cell size in fission yeast [7], mammalian [9,10], and nematode [9] cells so that mRNA concentrations stay roughly constant. Larger cells have more mRNA due to increased transcriptional output [59] as mRNA turnover rates are largely independent of cell size [7,9]. These observations raise the question of how cell size increases transcription rates despite the constant concentration of most proteins and the constant amount of DNA template outside S phase.

One likely explanation for how cell size promotes transcription is that the total transcriptional capacity of a cell is determined by the amount of transcriptional machinery rather than its concentration [9,60]. This could be the case if the transcriptional machinery is mostly bound to DNA and not freely diffusing in the nucleoplasm; that is, the transcriptional machinery is saturated. The biosynthetic capacity of a cell would then increase in proportion with cell size simply because larger cells have proportionally larger amounts of transcriptional machinery. However, eventually the genome itself becomes saturated, which is consistent with the observation that mutant yeast cells deficient in cell division exhibit a plateau in transcriptional output when grown to artificially large sizes [7]. In essence, the limiting-machinery model requires that transcription rates be determined by a currently unknown factor whose amount is proportional to cell size. While the identity of the limiting machinery is unknown, one candidate could be RNA polymerase. Not only does RNA polymerase II transcription increase with cell size in mammalian cells [9] but so does the amount of polymerase bound to DNA in yeast [7]. We note here that while our reference to experiments in yeast and animal cells presupposes a mechanism common to the two, this may not be the case.

The idea that transcription rates are limited by machinery rather than the template is further supported by the fact that protein and mRNA concentrations are independent of ploidy for most genes. That genome-wide transcription rates are limited by machinery is also supported by the observation that protein concentrations are independent of ploidy for most genes [61,62]. While ploidy has a large effect on average cell size (e.g., diploids are approximately twice as large as haploids in a given species), ploidy does not affect total RNA concentration or the relative mRNA levels of most genes [61,63,64]. That mRNA concentration is independent of ploidy is consistent with the transcriptional machinery being limiting rather than the gene template. However, the effect of ploidy on cell size complicates the interpretation. A crucial test of this model is the comparison of haploid and diploid cells. For cells of the same size, diploids spread the same amount of transcriptional machinery over twice the genomic template so that the transcription rate per gene is halved while the total transcription rate remains the same (Figure 3A). Consistent with this model, experiments show that Cln3 protein synthesis was the same in similarly sized haploid and diploid cells [35]. Moreover, a series of elegant mammalian cell fusion experiments revealed that when two differently sized mammalian cells are fused together, transcription of a single GFP allele is proportional to the amount of cytoplasm divided by the number of copies of the genome [9].

Figure 3. Origins of Size-Dependent and Size-Independent Protein Synthesis.

(A) The transcription rate of most genes is proportional to cell size. This is consistent with a model where the transcriptional machinery is limiting. Larger cells have more machinery, which results in a higher total rate of transcription. In the case where a haploid and a diploid cell are the same size, the total transcription rate will be the same because in the diploid the machinery is split among twice the number of templates. (B) In contrast to the general case described in (A), the budding yeast cell cycle inhibitor Whi5 is synthesized at a rate independent of cell size. This suggests that Whi5 synthesis is not limited by the transcriptional machinery and is simply proportional to the number of copies of the gene. Indeed, in the case where a haploid and a diploid cell are the same size, Whi5 is synthesized at twice the rate in the diploid.

The scaling of protein synthesis with cell size requires that in addition to transcription, also translation increase in proportion to cell size. One explanation for size-dependent increases in translation would be if the number of ribosomes increases in proportion to cell size, like most proteins. Consistent with this picture, rRNA and ribosomal protein synthesis increases with cell size [9,65] and the concentration of many ribosomal proteins is independent of cell size(D. Chandler-Brown, personal communication). Thus, if both mRNA and ribosome concentrations are independent of cell size, the protein synthesis rate per unit volume will also be independent of cell size, so that the total translation in the cell is directly proportional to its volume. The notion that protein synthesis rates are a large determinant of cell growth rate lends further support to growth rates being nearly proportional to cell size (i.e., approximately exponential growth at the level of single cells). While controversial, especially in the case of fission yeast [66], it is relatively clear that bacteria, budding yeast, and animal cells increase growth rates with cell size [65,67–72]. We note that cell size is not the only determinant of growth rate as similarly sized cells will exhibit significant variations in growth rate, which has been proposed to be due to variation in mitochondrial content in some cases [34,69,73,74]. In addition to the effects of cell size, changes in nutrient conditions also affect global protein synthesis rates. However, the relative synthesis rates of different proteins are usually maintained [58]. This is consistent with average synthesis rates being determined by a global limiting biosynthetic machinery that increases with cell size and good nutrient conditions, with relative differences between genes being determined by specific regulatory factors.

Breakdown of Scaling and its Biological Consequences

While the mechanisms described above show how larger cells can maintain constant molecular concentrations, they raise the question of how cells can produce size-dependent biochemical signals such as those triggering division. In budding yeast, the cell cycle inhibitor Whi5 is produced in cell size-independent amounts so that larger cells have lower concentrations and enter the cell cycle after a shorter period of cell growth.

In addition to its unusual size-independent synthesis, Whi5 also deviates in its response to changes in cell ploidy. Based on the limiting-machinery model, proteins can be expected to behave like the cell cycle activator Cln3, which is produced in proportion to cell size in both haploid and diploid cells (Figure 3A). Importantly, haploid and diploid cells that are the same size produce Cln3 at the same rate [35]. However, this is not the case for Whi5 [35]. Diploid cells produce Whi5 at roughly twice the size-independent rate of haploids even when they are the same size and therefore can be expected to have the same amount of limiting biosynthetic machinery (Figure 3B). The hemizygous WHI5/whi5Δ diploid synthesizes Whi5 at a similar rate to the haploid, suggesting that Whi5 synthesis is determined by the number of gene copies but is independent of ploidy and cell size. Moreover, it strongly suggests that the size independence of Whi5 synthesis is an intrinsic property of the gene itself and does not rely on feedback mechanisms monitoring Whi5 concentration.

While it is currently clear that size-independent Whi5 synthesis is highly relevant for cell size control, we do not yet know the underlying molecular mechanisms. Moreover, we also do not know how widespread this behavior may be across the genome nor whether size-dependent concentrations arise exclusively from synthesis-based mechanisms. One could just as easily imagine differential size dependencies in protein or mRNA degradation rates, particularly if a component of the degradation machinery is not produced proportionally to cell size. Certainly, many biological processes scaling with surface-to-volume ratios, like membrane synthesis, or scaling with genome length, like DNA synthesis, may be more efficient if smaller cells had higher concentrations of these enzymes. In addition, it might also be expected that proteins that are tightly bound to DNA, such as histones or high-affinity transcription factors, should be maintained at a constant number, not concentration, to match the constant DNA content.

Concluding Remarks

The establishment of the Whi5-dilution model represents a conceptual breakthrough in the field of cell size control because it shows how growth can impact any part of a regulatory network. Size-dependent Whi5 sits at the center of the budding yeast G1 control network whereas the size-independent Cln3 is upstream. Thus, unlike other biological signals, which originate at upstream components like receptors, growth can differentially affect the concentration of any signaling molecule to impact pathway activity (Figure 4).

Figure 4. Cell Growth Can Impact Any Component of a Regulatory Network via Specific Size Dependencies in Protein Synthesis.

Schematic illustration of the regulatory network controlling the budding yeast G1/S transition that links cell growth to division. Cell size impacts network activity through the concentration of the central cell cycle inhibitor Whi5, which is diluted by cell growth.

Here we have focused on how cell size controls division via the cell cycle control network. However, the specific architecture of the cell cycle control network is unlikely to be required for size-dependent cell cycle progression. All that is required would be the differential size dependency of the synthesis of proteins promoting and inhibiting network activity. Thus, differential size dependencies of regulatory protein synthesis can be used to encode cell size into the activity of any regulatory network.

Size-dependent protein synthesis can also be used to differentially coordinate organelle and cell size. In general, organelle size is determined by cell size through its influence on organelle subunit amounts. If, as proposed, organelle synthesis depletes a cellular pool of diffusible subunits, steady-state organelle size would be set by the total amount of these subunits [2]. The observation that organelle size typically is proportional to cell size is likely to result from the fact that most protein amounts, including organelle subunits, are proportional to cell size. However, while most protein amounts scale with cell size some do not, and this could be harnessed to differentially regulate organelle size. For example, it was recently shown that Caenorhabditis elegans embryos might be subject to this type of regulation. Larger and smaller embryos inherit a similar amount of a nucleolar protein, which is likely to account for the loss of proportionality between cell and nucleolar size [16]. This work highlights the potential for the differential scaling of protein synthesis to regulate complex cellular processes such as building organelles.

Despite the emerging understanding that differential protein synthesis has important biological functions including allowing the cell to generate a size-dependent biochemical signal, we do not understand the molecular mechanisms underlying the different size dependencies of protein synthesis (see Outstanding Questions). Further studies identifying these mechanisms will be likely to require quantitative cell biology approaches. While many biological signals are switch-like and can easily be discerned without quantification, other signals are not. In particular, signals reflecting continuous variables such as cell growth rate or cell size would be expected to change only twofold through a division cycle. Understanding such continuous processes requires careful quantitative analysis because such signals will appear nearly constant when examined by eye. For example, the basis of budding yeast cell size control could not have been identified without these methods. Two particular techniques stand out as essential: (i) the semiautomated segmentation, tracking, and analysis of single cells [75]; and (ii) the sufficiently accurate measurement of fluorescent protein concentration dynamics in single cells [35]. Given these two capabilities, single-cell analysis then allows natural cell-to-cell variation to be harnessed to distinguish between regulatory mechanisms. We anticipate that further studies will reveal the molecular mechanisms underlying size-dependent protein synthesis and reveal the extent to which cells use this differential size dependency to link diverse cellular functions with cell size and growth.

Outstanding Questions.

How does cell size determine biosynthesis? As a prerequisite to understand how cell size differentially affects the concentration of a specific subset of proteins, it is necessary to understand how cells globally coordinate biosynthesis to keep most proteins at a constant concentration.

Which proteins exhibit distinct size-dependent concentrations? As a next step, proteins that do not accumulate in proportion with cell size need to be identified. This will shed light on how cell size impacts diverse biological processes.

What molecular mechanisms determine the cell size dependence of protein concentrations? After identification of the set of differentially regulated proteins, transcription, translation, and degradation rates for these proteins have to be measured to reveal the underlying regulatory mechanisms.

Are the specific size dependencies of protein concentrations conserved? Basic biochemical requirements may impose a need for distinct size dependencies of specific protein concentrations that is similar for all eukaryotes. More work is required to see whether the identity of the differentially regulated proteins and the underlying molecular mechanisms are conserved.

Trends.

Whereas most macromolecules accumulate in proportion to cell size, some do not. The resulting differential size dependencies of molecular concentrations can be used to create size-dependent signals.

In budding yeast, the differential size dependence of the synthesis of the cell cycle activator Cln3 and the cell cycle inhibitor Whi5 results in cell size control.

In frog embryos, the differential size dependence of DNA and histone concentrations controls the mid-blastula transition of early development.